Backfilling Bond Prices with Changes in Option-Adjusted Spreads

Summary

The document discusses reconstructing historical bond prices for backtesting when a bond’s price history is incomplete. The proposed starting method regresses yield to maturity (YTM) on a benchmark, estimates missing yields from that relationship, and converts them into prices. The questioner reports a large discontinuity where the reconstructed series meets observed data, illustrating a limitation of fitting yield levels directly and then repricing the bond.

The answer recommends using option-adjusted spread (OAS), since the risk-free component of YTM is known and can be incorporated directly. It suggests regressing changes in the bond’s OAS on changes in a benchmark’s OAS, then building a fitted OAS series from those changes. This change-based construction is intended to avoid a level jump at the start of the backfill. The exchange offers a practical modeling suggestion, but it provides no empirical comparison, implementation details, or guidance on benchmark selection and model validation.

Key ideas

• YTM includes a known risk-free component that should be imposed rather than estimated through a benchmark regression.
• Modeling changes in OAS against changes in benchmark OAS can reduce discontinuities at the start of a backfilled series.
• A reconstructed spread series can be converted into estimated bond prices for backtesting.
• The suggested approach is conceptual and is not supported here by a backtest or comparison of alternative methods.

Tags

# How do I backfill the price of bonds for backtesting?


# How do I backfill the price of bonds for backtesting?












I need to backfill the price of bonds for testing a startegy.

The method employed is:

- Regressing the YTM of the bond against a benchmark.

- Using the regression estimates to calculate the YTM for the period in which the bond data is not available.

- Using the "price" function in excel to calculate the bond price with the earliest backfill date as the issue date.

However, using this method, there is a big difference in price at the point where my backfill data begins.

Could somebody please suggest any alternative?

## Answer by Tal Fishman (score 1)

https://quant.stackexchange.com/a/3679

First of all, you should be using OAS, not YTM, as the risk-free interest rate component of YTM is known and should be imposed rather than estimated. Second, rather than regressing OAS against a benchmark, regress changes in OAS against changes in a benchmark OAS. Then calculate a fitted OAS series from the change series, which naturally will not have any jumps where the backfill begins.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.